Update documentation occurrences of filename .sh
[gitweb.git] / commit.c
index 8248a994a50ab91874600567e544aa3735e4aa98..e8eb0aec5509a4fd09698bd15a37d1f4e8341f5e 100644 (file)
--- a/commit.c
+++ b/commit.c
@@ -9,6 +9,8 @@
 #include "gpg-interface.h"
 #include "mergesort.h"
 
+static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
+
 int save_commit_buffer = 1;
 
 const char *commit_type = "commit";
@@ -68,7 +70,7 @@ struct commit *lookup_commit_reference_by_name(const char *name)
        unsigned char sha1[20];
        struct commit *commit;
 
-       if (get_sha1(name, sha1))
+       if (get_sha1_committish(name, sha1))
                return NULL;
        commit = lookup_commit_reference(sha1);
        if (!commit || parse_commit(commit))
@@ -607,30 +609,17 @@ static struct commit *interesting(struct commit_list *list)
        return NULL;
 }
 
-static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
+/* all input commits in one and twos[] must have been parsed! */
+static struct commit_list *paint_down_to_common(struct commit *one, int n, struct commit **twos)
 {
        struct commit_list *list = NULL;
        struct commit_list *result = NULL;
        int i;
 
-       for (i = 0; i < n; i++) {
-               if (one == twos[i])
-                       /*
-                        * We do not mark this even with RESULT so we do not
-                        * have to clean it up.
-                        */
-                       return commit_list_insert(one, &result);
-       }
-
-       if (parse_commit(one))
-               return NULL;
-       for (i = 0; i < n; i++) {
-               if (parse_commit(twos[i]))
-                       return NULL;
-       }
-
        one->object.flags |= PARENT1;
        commit_list_insert_by_date(one, &list);
+       if (!n)
+               return list;
        for (i = 0; i < n; i++) {
                twos[i]->object.flags |= PARENT2;
                commit_list_insert_by_date(twos[i], &list);
@@ -669,9 +658,34 @@ static struct commit_list *merge_bases_many(struct commit *one, int n, struct co
                }
        }
 
-       /* Clean up the result to remove stale ones */
        free_commit_list(list);
-       list = result; result = NULL;
+       return result;
+}
+
+static struct commit_list *merge_bases_many(struct commit *one, int n, struct commit **twos)
+{
+       struct commit_list *list = NULL;
+       struct commit_list *result = NULL;
+       int i;
+
+       for (i = 0; i < n; i++) {
+               if (one == twos[i])
+                       /*
+                        * We do not mark this even with RESULT so we do not
+                        * have to clean it up.
+                        */
+                       return commit_list_insert(one, &result);
+       }
+
+       if (parse_commit(one))
+               return NULL;
+       for (i = 0; i < n; i++) {
+               if (parse_commit(twos[i]))
+                       return NULL;
+       }
+
+       list = paint_down_to_common(one, n, twos);
+
        while (list) {
                struct commit_list *next = list->next;
                if (!(list->item->object.flags & STALE))
@@ -709,6 +723,62 @@ struct commit_list *get_octopus_merge_bases(struct commit_list *in)
        return ret;
 }
 
+static int remove_redundant(struct commit **array, int cnt)
+{
+       /*
+        * Some commit in the array may be an ancestor of
+        * another commit.  Move such commit to the end of
+        * the array, and return the number of commits that
+        * are independent from each other.
+        */
+       struct commit **work;
+       unsigned char *redundant;
+       int *filled_index;
+       int i, j, filled;
+
+       work = xcalloc(cnt, sizeof(*work));
+       redundant = xcalloc(cnt, 1);
+       filled_index = xmalloc(sizeof(*filled_index) * (cnt - 1));
+
+       for (i = 0; i < cnt; i++)
+               parse_commit(array[i]);
+       for (i = 0; i < cnt; i++) {
+               struct commit_list *common;
+
+               if (redundant[i])
+                       continue;
+               for (j = filled = 0; j < cnt; j++) {
+                       if (i == j || redundant[j])
+                               continue;
+                       filled_index[filled] = j;
+                       work[filled++] = array[j];
+               }
+               common = paint_down_to_common(array[i], filled, work);
+               if (array[i]->object.flags & PARENT2)
+                       redundant[i] = 1;
+               for (j = 0; j < filled; j++)
+                       if (work[j]->object.flags & PARENT1)
+                               redundant[filled_index[j]] = 1;
+               clear_commit_marks(array[i], all_flags);
+               for (j = 0; j < filled; j++)
+                       clear_commit_marks(work[j], all_flags);
+               free_commit_list(common);
+       }
+
+       /* Now collect the result */
+       memcpy(work, array, sizeof(*array) * cnt);
+       for (i = filled = 0; i < cnt; i++)
+               if (!redundant[i])
+                       array[filled++] = work[i];
+       for (j = filled, i = 0; i < cnt; i++)
+               if (redundant[i])
+                       array[j++] = work[i];
+       free(work);
+       free(redundant);
+       free(filled_index);
+       return filled;
+}
+
 struct commit_list *get_merge_bases_many(struct commit *one,
                                         int n,
                                         struct commit **twos,
@@ -717,7 +787,7 @@ struct commit_list *get_merge_bases_many(struct commit *one,
        struct commit_list *list;
        struct commit **rslt;
        struct commit_list *result;
-       int cnt, i, j;
+       int cnt, i;
 
        result = merge_bases_many(one, n, twos);
        for (i = 0; i < n; i++) {
@@ -748,28 +818,11 @@ struct commit_list *get_merge_bases_many(struct commit *one,
        clear_commit_marks(one, all_flags);
        for (i = 0; i < n; i++)
                clear_commit_marks(twos[i], all_flags);
-       for (i = 0; i < cnt - 1; i++) {
-               for (j = i+1; j < cnt; j++) {
-                       if (!rslt[i] || !rslt[j])
-                               continue;
-                       result = merge_bases_many(rslt[i], 1, &rslt[j]);
-                       clear_commit_marks(rslt[i], all_flags);
-                       clear_commit_marks(rslt[j], all_flags);
-                       for (list = result; list; list = list->next) {
-                               if (rslt[i] == list->item)
-                                       rslt[i] = NULL;
-                               if (rslt[j] == list->item)
-                                       rslt[j] = NULL;
-                       }
-               }
-       }
 
-       /* Surviving ones in rslt[] are the independent results */
+       cnt = remove_redundant(rslt, cnt);
        result = NULL;
-       for (i = 0; i < cnt; i++) {
-               if (rslt[i])
-                       commit_list_insert_by_date(rslt[i], &result);
-       }
+       for (i = 0; i < cnt; i++)
+               commit_list_insert_by_date(rslt[i], &result);
        free(rslt);
        return result;
 }
@@ -780,6 +833,9 @@ struct commit_list *get_merge_bases(struct commit *one, struct commit *two,
        return get_merge_bases_many(one, 1, &two, cleanup);
 }
 
+/*
+ * Is "commit" a decendant of one of the elements on the "with_commit" list?
+ */
 int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
 {
        if (!with_commit)
@@ -789,28 +845,28 @@ int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
 
                other = with_commit->item;
                with_commit = with_commit->next;
-               if (in_merge_bases(other, &commit, 1))
+               if (in_merge_bases(other, commit))
                        return 1;
        }
        return 0;
 }
 
-int in_merge_bases(struct commit *commit, struct commit **reference, int num)
+/*
+ * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
+ */
+int in_merge_bases(struct commit *commit, struct commit *reference)
 {
-       struct commit_list *bases, *b;
+       struct commit_list *bases;
        int ret = 0;
 
-       if (num == 1)
-               bases = get_merge_bases(commit, *reference, 1);
-       else
-               die("not yet");
-       for (b = bases; b; b = b->next) {
-               if (!hashcmp(commit->object.sha1, b->item->object.sha1)) {
-                       ret = 1;
-                       break;
-               }
-       }
+       if (parse_commit(commit) || parse_commit(reference))
+               return ret;
 
+       bases = paint_down_to_common(commit, 1, &reference);
+       if (commit->object.flags & PARENT2)
+               ret = 1;
+       clear_commit_marks(commit, all_flags);
+       clear_commit_marks(reference, all_flags);
        free_commit_list(bases);
        return ret;
 }
@@ -819,51 +875,31 @@ struct commit_list *reduce_heads(struct commit_list *heads)
 {
        struct commit_list *p;
        struct commit_list *result = NULL, **tail = &result;
-       struct commit **other;
-       size_t num_head, num_other;
+       struct commit **array;
+       int num_head, i;
 
        if (!heads)
                return NULL;
 
-       /* Avoid unnecessary reallocations */
-       for (p = heads, num_head = 0; p; p = p->next)
-               num_head++;
-       other = xcalloc(sizeof(*other), num_head);
-
-       /* For each commit, see if it can be reached by others */
-       for (p = heads; p; p = p->next) {
-               struct commit_list *q, *base;
-
-               /* Do we already have this in the result? */
-               for (q = result; q; q = q->next)
-                       if (p->item == q->item)
-                               break;
-               if (q)
+       /* Uniquify */
+       for (p = heads; p; p = p->next)
+               p->item->object.flags &= ~STALE;
+       for (p = heads, num_head = 0; p; p = p->next) {
+               if (p->item->object.flags & STALE)
                        continue;
-
-               num_other = 0;
-               for (q = heads; q; q = q->next) {
-                       if (p->item == q->item)
-                               continue;
-                       other[num_other++] = q->item;
+               p->item->object.flags |= STALE;
+               num_head++;
+       }
+       array = xcalloc(sizeof(*array), num_head);
+       for (p = heads, i = 0; p; p = p->next) {
+               if (p->item->object.flags & STALE) {
+                       array[i++] = p->item;
+                       p->item->object.flags &= ~STALE;
                }
-               if (num_other)
-                       base = get_merge_bases_many(p->item, num_other, other, 1);
-               else
-                       base = NULL;
-               /*
-                * If p->item does not have anything common with other
-                * commits, there won't be any merge base.  If it is
-                * reachable from some of the others, p->item will be
-                * the merge base.  If its history is connected with
-                * others, but p->item is not reachable by others, we
-                * will get something other than p->item back.
-                */
-               if (!base || (base->item != p->item))
-                       tail = &(commit_list_insert(p->item, tail)->next);
-               free_commit_list(base);
        }
-       free(other);
+       num_head = remove_redundant(array, num_head);
+       for (i = 0; i < num_head; i++)
+               tail = &commit_list_insert(array[i], tail)->next;
        return result;
 }
 
@@ -1044,8 +1080,9 @@ static int excluded_header_field(const char *field, size_t len, const char **exc
        return 0;
 }
 
-struct commit_extra_header *read_commit_extra_header_lines(const char *buffer, size_t size,
-                                                          const char **exclude)
+static struct commit_extra_header *read_commit_extra_header_lines(
+       const char *buffer, size_t size,
+       const char **exclude)
 {
        struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
        const char *line, *next, *eof, *eob;
@@ -1112,8 +1149,92 @@ int commit_tree(const struct strbuf *msg, unsigned char *tree,
        return result;
 }
 
+static int find_invalid_utf8(const char *buf, int len)
+{
+       int offset = 0;
+
+       while (len) {
+               unsigned char c = *buf++;
+               int bytes, bad_offset;
+
+               len--;
+               offset++;
+
+               /* Simple US-ASCII? No worries. */
+               if (c < 0x80)
+                       continue;
+
+               bad_offset = offset-1;
+
+               /*
+                * Count how many more high bits set: that's how
+                * many more bytes this sequence should have.
+                */
+               bytes = 0;
+               while (c & 0x40) {
+                       c <<= 1;
+                       bytes++;
+               }
+
+               /* Must be between 1 and 5 more bytes */
+               if (bytes < 1 || bytes > 5)
+                       return bad_offset;
+
+               /* Do we *have* that many bytes? */
+               if (len < bytes)
+                       return bad_offset;
+
+               offset += bytes;
+               len -= bytes;
+
+               /* And verify that they are good continuation bytes */
+               do {
+                       if ((*buf++ & 0xc0) != 0x80)
+                               return bad_offset;
+               } while (--bytes);
+
+               /* We could/should check the value and length here too */
+       }
+       return -1;
+}
+
+/*
+ * This verifies that the buffer is in proper utf8 format.
+ *
+ * If it isn't, it assumes any non-utf8 characters are Latin1,
+ * and does the conversion.
+ *
+ * Fixme: we should probably also disallow overlong forms and
+ * invalid characters. But we don't do that currently.
+ */
+static int verify_utf8(struct strbuf *buf)
+{
+       int ok = 1;
+       long pos = 0;
+
+       for (;;) {
+               int bad;
+               unsigned char c;
+               unsigned char replace[2];
+
+               bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
+               if (bad < 0)
+                       return ok;
+               pos += bad;
+               ok = 0;
+               c = buf->buf[pos];
+               strbuf_remove(buf, pos, 1);
+
+               /* We know 'c' must be in the range 128-255 */
+               replace[0] = 0xc0 + (c >> 6);
+               replace[1] = 0x80 + (c & 0x3f);
+               strbuf_insert(buf, pos, replace, 2);
+               pos += 2;
+       }
+}
+
 static const char commit_utf8_warn[] =
-"Warning: commit message does not conform to UTF-8.\n"
+"Warning: commit message did not conform to UTF-8.\n"
 "You may want to amend it after fixing the message, or set the config\n"
 "variable i18n.commitencoding to the encoding your project uses.\n";
 
@@ -1170,7 +1291,7 @@ int commit_tree_extended(const struct strbuf *msg, unsigned char *tree,
        strbuf_addbuf(&buffer, msg);
 
        /* And check the encoding */
-       if (encoding_is_utf8 && !is_utf8(buffer.buf))
+       if (encoding_is_utf8 && !verify_utf8(&buffer))
                fprintf(stderr, commit_utf8_warn);
 
        if (sign_commit && do_sign_commit(&buffer, sign_commit))
@@ -1226,3 +1347,13 @@ struct commit_list **commit_list_append(struct commit *commit,
        new->next = NULL;
        return &new->next;
 }
+
+void print_commit_list(struct commit_list *list,
+                      const char *format_cur,
+                      const char *format_last)
+{
+       for ( ; list; list = list->next) {
+               const char *format = list->next ? format_cur : format_last;
+               printf(format, sha1_to_hex(list->item->object.sha1));
+       }
+}